Research and Development


This is why we get up in the morning...yes, really!

We are passionate about research and development!  Kije Sipi has continually and heavily invested in Hydromet Operational Intelligent (HMOI) R&D, developing its Weather Information System (WIS) and in the process establishing several innovative and powerful in-house HydroMet analytics programs to support our clients. As a result of this significant R&D commitment. the group has carved a unique market position as a niche HydroMet and water resources product and service provider in Canada and abroad. In particular, the following areas of research stand out as significant internal contributions to the list of corporate assets:

Real-time calibration of weather radar data.

Kije Sipi‘s research team is currently active in pursuing  real-time weather radar calibration technology in conjunction with its European academic partner. Preliminary research results are paving the way for the development of a cost-effective system that will dramatically improve NowCasting and post-event weather radar precipitation measurements.

Ground-level air temperature forecasts

The aim of this research project was to improve Environment and Climatic Change Canada’s ground-level air temperature forecasts throughout the entire continental track system of a large Canadian Class A railway client. The effort resulted in the 2019 deployment of a corrective prediction system that is heavily relied upon by operations staff. Big data mining and analyses of long-term historical meteorological forecasts and observed data were used to generate strongly correlated functions to adjust temperature forecasts. Surprising continental error trends were detected and effectively compensated.

Rainfall storm profiling & characterization

A set of specialized data mining programs was developed in-house to identify rainfall storm events within a huge weather radar database containing 15 years of weather radar data and resulting in more than 55 Trillion unique data points. Hydrometeorological criteria, including inter-event time and interspatial distance, were used in order to identify distinct storm events. Each of these precipitation events were subsequently characterized in terms of their magnitude (volume and intensity), frequency, duration, speed of travel and direction, return period as well as spatial decay statistics. This big data mining research led to  amassing quantitative rainfall storm statistics for thousands of events that ultimately yielded high resolution rainfall storm event characterizations and profiles.

National rainfall frequency statistics

Engineers and planners require rainfall frequency statistics when designing and planning water resources related structures. Specifically, there’s a need for a contiguous national coverage of rainfall frequency statistics (i.e. return periods) as opposed to the point Intensity-Duration-Frequency (IDF) data that is currently available. This set the stage for one of the first corporate R&D projects. The challenge here was to efficiently transform point-based data into a flexible and precise formulation for the continuous geographical mapping of storm magnitudes irrespective of location and storm duration. Using rainfall frequency statistics from over 500 meteorological stations in Canada and the USA, a set of four-parameters were resolved and transposed geographically in order to enable rainfall frequency computations for any radar grid cell anywhere in Canada…a unique capability.

An alternative rainfall design storm

The previously described in-house research on characterizing rainfall storm events further evolved in the development of a totally new method of designing Synthetic Design Storms for water resources engineering design applications. The proposed alternative to derive design storms is a major departure from the current use of point rainfall IDF data with synthetic rainfall intensity distributions. The new approach incorporate totally new concepts that are anchored in analyzing thousands of high resolution radar storm events that yielded the following innovative design elements:

  • Usage of rainfall statistics from analyzing the storm cell’s central (maximum) rainfall.
  • Usage of a 3D (X, Y, time) radar derived design storm according to duration.
  • Usage of adjustments for multiple storms cells within a watershed.

The unexpected research outcome of this project and eventual seminal work makes a strong case to pursue applied research. The results were published in the Canadian Meteorological & Oceanographic Society’s Bulletins.

Modelling of a national antecedent precipitation index (API)

A contiguous national Antecedent Precipitation Index (API) model was developed to simulate variations in soil moisture across Canada using weather radar as input.  Regionalization of the underlying parameters, such as soil properties, were completed in establishing a system for real time client applications.